TY  - CONF
AU  - Doncevic, Danimir
AU  - Schweidtmann, Artur M.
AU  - Vaupel, Yannic
AU  - Schäfer, Pascal
AU  - Caspari, Adrian
AU  - Mitsos, Alexander
TI  - Deterministic Global Nonlinear Model Predictive Control with Neural Networks Embedded
JO  - IFAC-PapersOnLine
VL  - 53
IS  - 2
SN  - 2405-8963
CY  - Laxenburg
PB  - IFAC
M1  - FZJ-2021-03179
SP  - 5273 - 5278
PY  - 2020
AB  - Nonlinear model predictive control requires the solution of nonlinear programs with potentially multiple local solutions. Here, deterministic global optimization can guarantee to find a global optimum. However, its application is currently severely limited by computational cost and requires further developments in problem formulation, optimization solvers, and computing architectures. In this work, we propose a reduced-space formulation for the global optimization of problems with recurrent neural networks (RNN) embedded, based on our recent work on feed-forward artificial neural networks embedded. The method reduces the dimensionality of the optimization problem significantly, lowering the computational cost. We implement the NMPC problem in our open-source solver MAiNGO and solve it using parallel computing on 40 cores. We demonstrate real-time capability for the illustrative van de Vusse CSTR case study. We further propose two alternatives to reduce computational time: i) reformulate the RNN model by exposing a selected state variable to the optimizer; ii) replace the RNN with a neural multi-model. In our numerical case studies each proposal results in a reduction of computational time by an order of magnitude.
T2  - 1st Virtual IFAC World Congress
CY  - 11 Jul 2020 - 17 Jul 2020, online (Germany)
Y2  - 11 Jul 2020 - 17 Jul 2020
M2  - online, Germany
LB  - PUB:(DE-HGF)16 ; PUB:(DE-HGF)8
UR  - <Go to ISI:>//WOS:000652593000151
DO  - DOI:10.1016/j.ifacol.2020.12.1207
UR  - https://juser.fz-juelich.de/record/894319
ER  -